doi: 10.2522/ptj.20150030Originally published online December 4, 2015

2016; 96:241-251.PHYS THER. Kenneth J. OttenbacherGraham, Alai Tan, Mukaila Raji, Carl V. Granger and Rebecca V. Galloway, Amol M. Karmarkar, James E.

DebilityInpatient Rehabilitation for Older Adults With Hospital Readmission Following Discharge From

http://ptjournal.apta.org/content/96/2/241found online at: The online version of this article, along with updated information and services, can be

Collections

Professional Issues     Health Services Research Special Series    

Health Services Research     Health Care System    

Geriatrics: Other     in the following collection(s): This article, along with others on similar topics, appears

e-Letters

"Responses" in the online version of this article. "Submit a response" in the right-hand menu under

or click onhere To submit an e-Letter on this article, click

E-mail alerts to receive free e-mail alerts hereSign up

by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

Hospital Readmission FollowingDischarge From InpatientRehabilitation for Older AdultsWith DebilityRebecca V. Galloway, Amol M. Karmarkar, James E. Graham, Alai Tan,Mukaila Raji, Carl V. Granger, Kenneth J. Ottenbacher

Background. Debility accounts for 10% of inpatient rehabilitation cases among Medicarebeneficiaries. Debility has the highest 30-day readmission rate among 6 impairment groupsmost commonly admitted to inpatient rehabilitation.

Objective. The purpose of this study was to examine rates, temporal distribution, andfactors associated with hospital readmission for patients with debility up to 90 days followingdischarge from inpatient rehabilitation.

Design. A retrospective cohort study was conducted using records for 45,424 Medicarefee-for-service beneficiaries with debility discharged to community from 1,199 facilities during2006–2009.

Methods. Cox proportional hazard regression models were used to estimate hazard ratiosfor readmission. Schoenfeld residuals were examined to identify covariate-time interactions.Factor-time interactions were included in the full model for Functional Independence Measure(FIM) discharge motor functional status, comorbidity tier, and chronic pulmonary disease.Most prevalent reasons for readmission were summarized by Medicare severity diagnosisrelated groups.

Results. Hospital readmission rates for patients with debility were 19% for 30 days and 34%for 90 days. The highest readmission count occurred on day 3 after discharge, and 56% ofreadmissions occurred within 30 days. A higher FIM discharge motor rating was associatedwith lower hazard for readmissions prior to 60 days (30-day hazard ratio�0.987; 95% confi-dence interval�0.986, 0.989). Comorbidities with hazard ratios �1.0 included comorbiditytier and 11 Elixhauser conditions, 3 of which (heart failure, renal failure, and chronic pulmo-nary disease) were among the most prevalent reasons for readmission.

Limitations. Analysis of Medicare data permitted only use of variables reported for admin-istrative purposes. Comorbidity data were analyzed only for inpatient diagnoses.

Conclusions. One-third of patients were readmitted to acute hospitals within 90 daysfollowing rehabilitation for debility. Protective effect of greater motor function was diminishedby 60 days after discharge from inpatient rehabilitation.

R.V. Galloway, PT, PhD, Depart-ment of Physical Therapy, Univer-sity of Texas Medical Branch, 301University Blvd, Galveston, TX77555-1144 (USA). Address allcorrespondence to Dr Gallowayat: regallow@utmb.edu.

A.M. Karmarkar, PhD, MPH, Divi-sion of Rehabilitation Sciences,University of Texas MedicalBranch.

J.E. Graham, PhD, DC, Division ofRehabilitation Sciences, Universityof Texas Medical Branch.

A. Tan, MD, PhD, Institute forTranslational Sciences, Universityof Texas Medical Branch.

M. Raji, MD, MS, Department ofInternal Medicine, Geriatrics, Uni-versity of Texas Medical Branch.

C.V. Granger, MD, Uniform DataSystem for Medical Rehabilitation,UB Foundation Activities Inc, Buf-falo, New York.

K.J. Ottenbacher, PhD, OTR, Divi-sion of Rehabilitation Sciences,University of Texas MedicalBranch.

[Galloway RV, Karmarkar AM, Gra-ham JE, et al. Hospital readmissionfollowing discharge from inpa-tient rehabilitation for older adultswith debility. Phys Ther. 2016;96:241–251.]

© 2016 American Physical TherapyAssociation

Published Ahead of Print:December 4, 2015

Accepted: November 25, 2015Submitted: January 20, 2015

Health Services ResearchSpecial Series

Post a Rapid Response tothis article at:ptjournal.apta.org

February 2016 Volume 96 Number 2 Physical Therapy f 241 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

Debility, also known as decondi-tioning, involves decline in func-tional mobility or activities of

daily living, or both.1,2 This conditioncommonly occurs in older adults duringan acute hospitalization, with profounddecline, such as loss of independence,observed in some patients.3–5 For exam-ple, a multicenter trial showed that 1 in 6previously independent older adults wasunable to independently walk across aroom after an acute hospital stay.5 Refer-ral to a postacute inpatient rehabilitationfacility may be indicated for patients whoexperience acute functional decline asso-ciated with debility and demonstratepotential to improve from intensiveinterdisciplinary physical rehabilitation.6

Debility as an admission diagnosis forinpatient rehabilitation is defined as gen-eralized deconditioning not attributableto any other Centers for Medicare &Medicaid Services (CMS) rehabilitationimpairment groups, such as stroke ororthopedic conditions.7 Inpatient reha-bilitation is the only postacute settingthat has an admission group specificallyfor debility. Primary medical diagnosisfor this impairment group may be debil-ity, generalized weakness, or an infectionor other multisystem pathology forwhich there are numerous diagnosticcodes.8

The percentage of Medicare beneficia-ries with debility receiving postacuterehabilitation increased substantially inthe past decade.9 In 2004, 6% of benefi-ciaries receiving inpatient medical reha-bilitation were in the debility rehabilita-tion impairment group.9 Between 2010and 2013, 10% of beneficiaries in inpa-tient rehabilitation were admitted fordebility.9 Patients with debility representthe fourth largest impairment groupreceiving inpatient rehabilitation ser-vices following stroke, other neurologi-cal conditions (eg, multiple sclerosis),and lower extremity fracture.9 There isrelatively little research examining out-comes for patients with debility.

A critical outcome for inpatient rehabili-tation is the patient’s ability to return tocommunity living. The debility impair-ment group has a high prevalence (11%–14%) of discharges directly from inpa-

tient rehabilitation to acute care settingscompared with other impairmentgroups.2,8,10,11 Patients who are dis-charged to a community setting are alsoat risk of readmission to an acute caresetting.12 A recent study of hospital read-mission for the 6 largest inpatient reha-bilitation impairment categories showedthat debility had the highest rate, with19% of patients readmitted to acute carehospitals within 30 days of discharge tothe community.12 Frequent reasons forreadmission included heart failure, septi-cemia, and kidney and urinary tractinfections.12

The CMS recently proposed 30-day read-mission to acute care hospitals followingdischarge from inpatient rehabilitation asa quality measure to begin in fiscal year201713; this measure has been endorsedby the National Quality Forum.14 Thus,research to better understand patientand facility factors associated with highreadmission risk in people who havecommon conditions, such as debility, isimportant.

Existing research on inpatient rehabilita-tion for patients with debility hasfocused primarily on descriptive out-comes upon discharge2,8,10,11 or within30 days.12 Further investigation of factorsassociated with key postdischarge mea-sures such as readmission is merited toguide evidence-based clinical decisionmaking and strategies for prevention ofadverse outcomes. The objective of thisstudy was to investigate the rates, distri-bution, and factors associated with hos-pital readmission for up to 90 days afterdischarge from inpatient rehabilitationfor older adults receiving inpatient reha-bilitation for debility. Although 30 days iscurrently relevant to readmission as aquality measure for CMS, we followedpatients for 90 days based on proposedchanges in the delivery of postacutecare. For example, CMS’s Bundled Pay-ments for Care Improvement (BPCI) ini-tiative is currently studying services andpayment models based on episodes ofcare that may extend to 90 days.15 Weused Medicare claims files to examinethe temporal distribution of hospitalreadmission rates within 90 days afterdischarge and the associated factors for

older adults receiving inpatient rehabili-tation for debility.

MethodStudy SampleWe studied a sample of Medicare fee-for-service beneficiaries who received inpa-tient rehabilitation for debility from Jan-uary 2006 through September 2009. TheBeneficiary Summary File includes Medi-care beneficiary enrollment informa-tion,16 which we used to create an ana-lytical file based on inclusion andexclusion criteria. We used CMS In-patient Rehabilitation Facility–PatientAssessment Instrument (IRF-PAI) data toidentify debility impairment group,demographic variables, admission vari-ables, impairment group code, dischargeinformation, and functional status rat-ings.7 Medicare Provider Analysis andReview (MedPAR) files were used toidentify hospital readmissions and cor-mobidities.17 The research was reviewedby the University of Texas MedicalBranch Institutional Review Board andcomplied with the data use agreementobtained from CMS.

The debility impairment group codeincluded “cases with generalized decon-ditioning not attributable to any of theother Impairment Groups.”7 The Reha-bilitation Impairment Category (RIC) fordebility was “miscellaneous” (20).7 Theinitial sample contained 130,148 Medi-care beneficiaries identified as recipientsof inpatient rehabilitation during 2006–2009 in the CMS debility impairmentgroup. Additional inclusion criteriawere: (1) patient lived in the communityprior to hospitalization; (2) admissiondirectly from acute care for initial inpa-tient rehabilitation; (3) record includedno program interruptions (a programinterruption occurred when a patientwas temporarily transferred to an acutecare setting for up to 3 days and thenreturned for further inpatient rehabilita-tion7); (4) inpatient rehabilitation lengthof stay was between 3 and 30 days; (5)Medicare beneficiary did not reside inthe state of Maryland (different CMS pay-ment structure); (6) patient was dis-charged to a community setting by Sep-tember 30, 2009 (to allow for 90 days offollow-up after discharge); (7) age atadmission was between 66 and 100

Hospital Readmission and Older Adults With Debility

242 f Physical Therapy Volume 96 Number 2 February 2016 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

years; (8) Medicare entitlement was forOld Age and Survivors Insurance; (9)Medicare fee-for-service status; and (10)beneficiary did not die during inpatientrehabilitation or within 90 days postdis-charge. Of the 130,148 Medicare benefi-ciaries in the debility impairment group,45,424 (35%) met the criteria and wereincluded in the eligible sample. Figure 1illustrates the flow and number ofpatients included and excluded in eachphase of the data set construction. Thiscohort included patients from 1,199inpatient rehabilitation facilities.

Variable DefinitionsHospital readmission, to a short-stayacute care hospital, was the dependentoutcome identified using CMS MedPARinformation for 90 days after dischargefrom inpatient rehabilitation. Patientsdischarged directly from inpatient reha-bilitation to acute care hospital wereexcluded from this calculation. Time,measured in days, from inpatient rehabil-itation discharge to acute hospital read-mission was used in the analysis. Onlythe first readmission following dischargewas included.

Reasons for hospital readmission wereidentified by Medicare severity diagnosisrelated groups (MS-DRG). The MS-DRG isa classification system for hospital inpa-tient prospective payment systems thataccounts for illness severity and resourceutilization for Medicare beneficiaries.18

Levels of illness severity are: MCC(major complication/comorbidity), CC(complication/comorbidity), and non-complication/comorbidity.18 The MCCand CC illness severities for the samediagnosis were combined for descriptivepurposes.

Patient characteristics were extractedfrom the IRF-PAI file, including age (con-tinuous), marital status (married versusnot married), sex, race/ethnicity (non-Hispanic white, black, Hispanic, andother), and prehospital living status(alone versus with someone). Living inthe community prior to hospitalizationand community discharge were recodedfrom the prior living setting and dis-charge setting variables, respectively, inthe IRF-PAI files; community includedhome, assisted living residence, board

and care, and transitional living catego-ries in both variables.7,19

Functional status was determined usingthe 18 items originally developed for theFunctional Independence Measure (FIM)instrument19 and included in the IRF-PAI.The 18 items were divided into motorand cognitive subscales. Thirteen motoritems assessed self-care (eating, groom-

ing, bathing, upper body dressing, lowerbody dressing, and toileting), sphinctercontrol (bowel and bladder), transfers(bed/chair/wheelchair, toilet, and tub/shower), and locomotion (walk/wheel-chair and stairs).7,20,21 Five cognitiveitems assessed communication (compre-hension and expression) and social cog-nition (social interaction, problem solv-ing, and memory).7,20,21 Each item was

Figure 1.Flowchart for inclusion and exclusion criteria. IRF�inpatient rehabilitation facility, IRF-PAI�Inpatient Rehabilitation Facility–Patient Assessment Instrument, OASI�Old Age andSurvivors Insurance, FFS�fee-for-service, MedPAR�Medicare Provider Analysis and Review.

Hospital Readmission and Older Adults With Debility

February 2016 Volume 96 Number 2 Physical Therapy f 243 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

rated from 1 (“complete dependence”)to 7 (“complete independence”), for atotal possible rating of 18 to 126.7 Higherratings represented greater functionalindependence. Functional status itemswere administered by a trained profes-sional within 36 hours of admission anddischarge. The reliability, validity, andresponsiveness for the functional statusitems have been examined by CMS andothers and found to be adequate.22–29

Two measures of comorbidity wereexamined. The case-mix group (CMG)comorbidity tier was developed by CMSas part of the prospective payment sys-tem for inpatient rehabilitation. Thesetier categories were based on the pres-ence of specific comorbidities associatedwith increased costs.30 Payments were

adjusted for comorbidities using a 4-tiersystem: tier 1 (high cost), tier 2 (mediumcost), tier 3 (low cost), and no tier.31 If apatient had more than one comorbidityon a tier list, the highest cost tier wasassigned.7 Examples of tier 1 (high cost)comorbidities from 2006–2009 werevocal paralysis, tracheostomy, and renaldialysis.32 We also used the 29 Elixhausercomorbidities derived from MedPARdata using the Agency for HealthcareResearch and Quality (AHRQ)-Web ICD-9-CM version to assess conditions notlinked directly to payment.33 Elixhauseret al34 recommended analysis of eachcomorbidity variable rather than a sim-plified index for applicability to variousdiseases.

Data AnalysisTime to hospital readmission, measuredin days, after discharge from inpatientrehabilitation was the dependent out-come for time-to-event analysis. The log-rank test of survival function across 4years was used to validate the appropri-ateness of combining all 4 years. Patientswho did not experience readmission dur-ing 90 days following discharge frominpatient rehabilitation were censored asno readmission.35,36 We plotted actualreadmission day frequency counts as ahistogram over the entire 90-day studyperiod. We also calculated cumulativeunadjusted readmission rates for the fol-lowing day intervals: 3, 7, 15, 30, 60, and90 days. These time points correspond tothe first few days following discharge,

Table 1.Sample Characteristics Stratified by Hospital Readmission Status 90 Days After Discharge From Inpatient Rehabilitation for Debilitya

Variable Total

Readmission

PYes No

Patients, n 45,424 15,439 29,985

Readmission rate 34.0% 100% 0%

Age (y), X (SD) 80.8 (7.0) 80.6 (7.0) 80.9 (7.0) �.001

Sex (female) 60.1% 57.9% 61.2% �.001

Race/ethnicityb (n�44,461)

.003

White 86.4% 85.9 86.6%

Black 7.4% 8.0% 7.2%

Hispanic 3.0% 3.1% 2.9%

Other 1.1% 0.97% 1.2%

Marital statusb (n�44,891).002

Not married 56.2% 55.2% 56.7%

Living situationb (n�43,627)�.001

Living with others 61.8% 64.5% 60.4%

CMG comorbidity tierb (n�45,422)

�.001

No tier 57.2% 52.3% 59.7%

Low cost 29.1% 30.6% 28.3%

Medium cost 9.9% 11.7% 8.9%

High cost 3.9% 5.5% 3.1%

LOS (d), X (SD) 11.9 (4.7) 12.4 (4.8) 11.7 (4.6) �.001

FIM admission motor, X (SD) 41.4 (10.8) 40.4 (10.8) 42.0 (10.7) �.001

FIM admission cognition, X (SD) 25.3 (6.4) 24.9 (6.5) 25.4 (6.3) �.001

FIM discharge motor, X (SD) 65.0 (12.1) 63.0 (12.7) 66.0 (11.6) �.001

FIM discharge cognition, X (SD) 28.7 (5.2) 28.3 (5.4) 28.9 (5.1) �.001

a CMG�case-mix group, LOS�length of stay, FIM�Functional Independence Measure.b Missing values with sample size provided.

Hospital Readmission and Older Adults With Debility

244 f Physical Therapy Volume 96 Number 2 February 2016 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

the first week, midway to first month,and monthly for the study period.

Characteristics of patients who werereadmitted to an acute care hospital 90days following discharge from inpatientrehabilitation were compared to patientswho did not experience readmission.Missing data for race/ethnicity, maritalstatus, living situation, and CMG comor-bidity tier were reported in Table 1. Nodata were missing for functional status.Differences in readmission status wereevaluated with t tests for continuous vari-ables and chi-square tests for categoricalvariables. The Spearman correlationmatrix for CMG comorbidity tier andindividual Elixhauser comorbidities wasexamined for potential collinearity priorto multivariable modeling. The criterionfor strong association was a correlationcoefficient greater than .75.37

Hazard ratios for covariate parameterswere computed using Cox proportionalhazard regression. Covariates in the fullmodel included demographics, func-tional motor and cognitive subscales atdischarge, CMG comorbidity tier, andElixhauser comorbidities with significantbivariate differences. An assumption ofCox regression analysis was consistencyof hazard over time. Schoenfeld residualswere examined for potential interactionsbetween individual covariates andtime.36 Interaction terms were subse-quently included for factors that variedsignificantly with time (P�.05).

Statistical analyses were conducted usingSAS (version 9.2) LIFETEST36,38 andPHREG36,38,39 procedures. The aggregateoption was used in the model statementto account for facility as a cluster vari-able.35 The exact method was used forties with readmission time.36 Adjustedhazard ratios were computed with 95%confidence intervals (CIs). The depen-dent variable was hazard (risk)40 of hos-pital readmission after discharge frominpatient rehabilitation. Covariates withhazard ratios greater than 1.0 were asso-ciated with a higher probability ofhospital readmission after dischargefrom inpatient rehabilitation.40 A hazardratio less than 1.0 indicated that the cova-riate was protective, associated with

decreased probability of readmission.40

Level of statistical significance was .05.

Role of the Funding SourceThis study was supported, in part, bygrants from the National Institutesof Health (R24 HD065702 and R01-HD069443) and the National Institute onDisability and Rehabilitation Research(H133G080163).

ResultsDistribution of hospital readmission timeafter discharge from inpatient rehabilita-tion depicted the highest peak in fre-quency of readmission within the firstweek (Fig. 2). Cumulative readmissionrates were 2.9% for 3 days, 6.7% for 7days, 12.1% for 15 days, 18.9% for 30days, 27.7% for 60 days, and 34.0% for90 days after discharge from inpatientrehabilitation. For the 15,439 patientswho experienced hospital readmissionwithin 90 days, the mean time from inpa-tient rehabilitation discharge to readmis-sion was 32.4 days (SD�25.6). Fifty-sixpercent of readmissions occurred within30 days, and 82% of readmissionsoccurred within 60 days. Survival proba-bility was similar across discharge years2006–2009 (P�.24). Therefore, all yearswere combined in subsequent analyses.

Patient and ClinicalCharacteristicsThe mean patient age was 80.8 years(SD�7.0). The majority of patients were

female (60.1%), non-Hispanic white(86.4%), not married (56.2%), and livingwith others prior to acute hospitalization(61.8%). The mean length of inpatientrehabilitation stay was 11.9 days(SD�4.7). Mean FIM ratings at dischargewere 65.0 (SD�12.1) for motor subscale,28.7 (SD�5.2) for cognition subscale,and 93.7 (SD�15.3) for FIM total. Thesecharacteristics were stratified by read-mission outcome in Table 1. The FIMdischarge motor score was 3 pointslower (63.0 versus 66.0) for patientswho were readmitted. Percentages ofpatients with low-, medium-, and high-cost CMG comorbidity tiers were higherfor the readmission group.

Elixhauser comorbidities are listed byreadmission status in Table 2 in descend-ing order of prevalence. Seventeen con-ditions had significantly different fre-quencies among readmission groups:hypertension, chronic pulmonary dis-ease, congestive heart failure, fluid andelectrolyte disorders, renal failure, hypo-thyroidism, other neurological disorders,peripheral vascular disease, diabeteswith chronic complications, depression,weight loss, solid tumor without metas-tasis, obesity, coagulopathy, metastaticcancer, lymphoma, and liver disease. Theremaining 12 conditions were not signif-icantly different among readmissiongroups. Correlation coefficients betweenpairs of Elixhauser comorbidities hadlow strength and ranged from �.10 to

Figure 2.Distribution of hospital readmission for 90 days after discharge from inpatient rehabilitationfor debility.

Hospital Readmission and Older Adults With Debility

February 2016 Volume 96 Number 2 Physical Therapy f 245 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

.12. Correlation coefficients betweenCMG comorbidity tiers and individualElixhauser comorbidities were weak tofair (.11–.26) and thus did not indicateconcern for collinearity in the multivari-able model.

Multivariable Survival AnalysisCorrelations between Schoenfeld residu-als for covariates and time were signifi-cant for FIM discharge motor rating(P�.0001), CMG comorbidity tier–me-dium cost (P�.0001), chronic pulmo-

nary disease (P�.01), and fluid/electro-lyte disorders (P�.02). Interaction termsfor these covariates with readmissiontime were included in the Cox regres-sion model. Interaction for fluid/electro-lyte disorders with time was subse-quently dropped as nonsignificant(P�.82).

Hazard ratios, adjusted for all other cova-riates, for hospital readmission within 90days of discharge from inpatient rehabil-itation are depicted in Table 3. For vari-

ables with significant time interactions,hazard ratios were reported by readmis-sion time (days 3, 7, 15, 30, 60, and 90)(Tab. 3). Single hazard ratios werereported for all other variables, as hazarddid not significantly vary with time ofreadmission (Tab. 3). Age was associatedwith 0.5% decreased readmission hazardfor 1 year older. Sex was not a significantfactor, and race/ethnicity was only pro-tective for the other group (not black orHispanic) compared with white. Notmarried was associated with a 6%

Table 2.Elixhauser Comorbidities Stratified by Hospital Readmission Status 90 Days After Discharge From Inpatient Rehabilitation for Debility

Comorbidity Total

Readmission

PaYes No

Hypertension 56.5% 53.1% 58.3% �.0001

Chronic pulmonary disease 21.2% 23.1% 20.3% �.0001

Congestive heart failure 19.5% 23.3% 17.5% �.0001

Diabetes without chronic complications 19.1% 18.9% 19.2% .53

Deficiency anemias 17.6% 17.3% 17.8% .21

Fluid and electrolyte disorders 15.4% 16.0% 15.1% .02

Renal failure 15.1% 19.2% 12.9% �.0001

Hypothyroidism 10.6% 9.3% 11.3% �.0001

Other neurological disorders 9.4% 8.9% 9.6% .009

Peripheral vascular disease 9.3% 10.1% 8.8% �.0001

Diabetes with chronic complications 7.6% 8.8% 6.9% �.0001

Depression 7.5% 6.8% 7.9% �.0001

Weight loss 7.0% 7.6% 6.6% �.0001

Solid tumor without metastasis 6.5% 7.0% 6.3% .002

Paralysis 5.7% 5.7% 5.7% .87

Obesity 5.0% 4.5% 5.3% .0002

Valvular disease 4.8% 5.0% 4.7% .23

Rheumatoid arthritis/collagen vascular diseases 3.5% 3.6% 3.4% .24

Pulmonary circulation disease 3.0% 3.0% 2.9% .66

Coagulopathy 2.5% 3.1% 2.2% �.0001

Metastatic cancer 2.3% 3.4% 1.8% �.0001

Psychoses 2.1% 2.0% 2.2% .20

Lymphoma 1.6% 2.2% 1.2% �.0001

Chronic blood loss anemia 1.3% 1.3% 1.2% .41

Liver disease 1.1% 1.5% 0.86% �.0001

Alcohol abuse 0.88% 0.78% 0.93% .10

Drug abuse 0.11% 0.09% 0.11% .48

Peptic ulcer disease excluding bleeding 0.06% 0.06% 0.05% .69

Acquired immunodeficiency syndrome 0.02% 0.03% 0.02% .33

a Seventeen comorbidities with significant differences included in Cox regression.

Hospital Readmission and Older Adults With Debility

246 f Physical Therapy Volume 96 Number 2 February 2016 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

increase in readmission hazard, but livingalone versus with someone was notsignificant.

Higher FIM discharge motor ratings weremore protective for early hospital read-missions; hazard ratios incrementallyincreased to 1.0 (95% CI�0.999, 1.002)by day 60. Further analysis of individualFIM discharge motor items revealed sig-nificant adjusted hazard ratios (unit�1FIM point) for walk/wheelchair locomo-tion (0.95; 95% CI�0.93, 0.96), stairlocomotion (0.96; 95% CI�0.95, 0.97),lower body dressing (0.96; 95% CI�0.94,0.98), eating (0.96; 95% CI�0.94, 0.98),bowel control (0.97; 95% CI�0.96,0.99), and bladder control (0.98; 95%CI�0.97, 0.99). As observed for the FIMdischarge motor subscale (Tab. 3), blad-der control, walk/wheelchair locomo-tion, and stair locomotion were not pro-tective by day 60. Eating and bowelcontrol were not protective by day 30.Lower body dressing was the only signif-icant motor item that did not interactwith time (ie, remained protectivethrough day 90). The FIM dischargemotor items not associated with readmis-sion were grooming, bathing, upperbody dressing, toileting, bed/chair/wheelchair transfers, toilet transfers, andtub/shower transfers.

Comorbidities significantly associatedwith higher hazard of readmissionincluded: CMG comorbidity tier andchronic pulmonary disease (up to 30days), congestive heart failure, fluid/elec-trolyte disorders, renal failure, peripheralvascular disease, weight loss, solid tumorwithout metastasis, coagulopathy, meta-static cancer, lymphoma, and liver dis-eases. Comorbidities significantly associ-ated with lower hazard for readmission(hazard ratios �1.0) were hypertension,hypothyroidism, other neurological dis-orders, and obesity.

Causes of ReadmissionTable 4 lists the most prevalent reasonsfor hospital readmission as coded byMS-DRG. The cumulative percentage forthese 14 readmission categories was37%. Common diagnoses included heartfailure, kidney/urinary tract infections,renal failure, pneumonia, chronicobstructive pulmonary disease, nutri-

Table 3.Results of Cox Regression Model for Hospital Readmission Within 90 Days of DischargeFrom Inpatient Rehabilitation for Debilitya

Characteristics HR 95% CI

Age (unit�1 y) 0.995 0.992, 0.997

Sex (male vs female) 1.01 0.98, 1.05

Race/ethnicity

Black vs white 0.96 0.91, 1.02

Hispanic vs white 0.93 0.86, 1.01

Other vs white 0.83 0.73, 0.93

Not married vs married 1.06 1.02, 1.10

Living with someone vs alone 1.03 0.99, 1.07

FIM discharge cognitive (unit�1) 1.00 0.996, 1.002

FIM discharge motor (unit�1)b

Day 3 0.976 0.974, 0.977

Day 7 0.977 0.976, 0.979

Day 15 0.981 0.980, 0.982

Day 30 0.987 0.986, 0.989

Day 60 1.000 0.999, 1.002

Day 90 1.014 1.012, 1.015

CMG comorbidity tierb

Tier 3 (low cost) vs no tier

Day 3 1.14 1.08, 1.19

Day 7 1.13 1.08, 1.18

Day 15 1.11 1.06, 1.16

Day 30 1.08 1.04, 1.12

Day 60 1.02 0.98, 1.06

Day 90 0.96 0.90, 1.02

Tier 2 (medium cost) vs no tier

Day 3 1.30 1.22, 1.39

Day 7 1.29 1.21, 1.37

Day 15 1.26 1.19, 1.34

Day 30 1.22 1.16, 1.28

Day 60 1.14 1.08, 1.20

Day 90 1.07 0.98, 1.16

Tier 1 (high cost) vs no tier

Day 3 1.55 1.40, 1.72

Day 7 1.50 1.36, 1.65

Day 15 1.40 1.28, 1.53

Day 30 1.24 1.15, 1.33

Day 60 0.97 0.90, 1.04

Day 90 0.76 0.68, 0.83

Hypertension 0.94 0.91, 0.97

(Continued)

Hospital Readmission and Older Adults With Debility

February 2016 Volume 96 Number 2 Physical Therapy f 247 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

tional disorders, and gastrointestinal dis-orders. More than 500 different MS-DRGcodes were listed for readmissionswithin 90 days of discharge from inpa-tient rehabilitation for debility.

DiscussionWe examined rate of hospital readmis-sion and factors associated with hazardof readmission for 90 days following dis-charge from inpatient rehabilitation in alarge sample of older adults with debility.One-third of the cohort experiencedreadmission. This finding was consistentwith 90-day readmission rates amongMedicare fee-for-service beneficiariesafter acute hospitalization.41 The 30-dayhospital readmission rate of 19% in ourstudy was not as high as reported forMedicare beneficiaries in skilled nursingfacilities (23%–24%).42,43 However, ourstudy focused on patients who were dis-

charged to the community, with readmis-sion time beginning after discharge frominpatient rehabilitation rather than acutehospital. Frequency distribution showedthat the highest number of readmissionsoccurred on day 3 following dischargefrom inpatient rehabilitation. More thanhalf (56%) of readmissions in the 90-dayobservation period occurred within thefirst month. This information may help indeveloping postdischarge monitoringand treatment plans.

We studied hazard for demographic,functional, and comorbidity factors asso-ciated with hospital readmission. Demo-graphic factors associated with readmis-sion hazard for patients with debilitywere age, race/ethnicity (other versuswhite), and marital status. Older age as aprotective factor differed from findingsin other readmission studies.41,44 Survi-

vor effect was a potential influence ifolder patients with debility who survivedthe observation period were healthier.45

Comorbidity findings for patients withdebility indicated that higher-cost CMGcomorbidity tiers were associated withhigher readmission hazard up to 30 daysfollowing discharge. For Elixhausercomorbidities, congestive heart failure,renal failure, and chronic pulmonary dis-ease were among the top reasons associ-ated with hospital readmission. Thesegenerally prevalent conditions haveimplications for developing and target-ing hospital readmission reduction pro-grams for patients with debility. Forinstance, heart failure was also the mostfrequent reason for readmission after dis-charge from acute care for patients withMedicare fee-for-service41 and 1 of 3 diag-noses measured for the acute care hos-pital readmissions reduction program.46

Metastatic cancer was a less prevalentcondition but had the highest hazardratio among the Elixhauser comorbidi-ties. Weight change was also a notewor-thy factor, as weight loss, weakness, andlow physical activity are operational cri-teria for frailty.47 The relationshipbetween frailty and debility is a poten-tially valuable topic for future research.

Of the Elixhauser comorbidities withhazard ratios less than 1, paradoxicaleffects have been described in literaturefor hypertension48 and obesity.49 Inpatients with heart failure, low bloodpressure was associated with increasedrisk of long-term mortality and hospital-ization.48 Although obesity is a risk factorfor development of heart failure, it hasbeen associated with lower risk-adjustedmortality in patients with establishedheart failure.49 A theory applicable topatients with debility and comorbiditiesis that greater adipose tissue providesreserve for catabolic changes that occurwith disease processes.49

Functional status was important to exam-ine because it is a primary outcome mea-sure for inpatient rehabilitation and apotentially modifiable factor. For olderadults with debility, FIM discharge motorsubscale ratings, but not cognition sub-scale ratings, were associated with lowerreadmission. This finding was consistent

Table 3Continued

Characteristics HR 95% CI

Chronic pulmonary diseaseb

Day 3 1.15 1.10, 1.21

Day 7 1.14 1.09, 1.19

Day 15 1.12 1.07, 1.17

Day 30 1.08 1.04, 1.12

Day 60 1.00 0.97, 1.04

Day 90 0.93 0.89, 0.99

Congestive heart failure 1.16 1.12, 1.21

Fluid and electrolyte disorders 1.07 1.03, 1.12

Renal failure 1.22 1.17, 1.27

Hypothyroidism 0.94 0.90, 0.99

Other neurological disorders 0.92 0.87, 0.97

Peripheral vascular disease 1.12 1.07, 1.18

Diabetes with chronic complications 1.04 0.98, 1.11

Depression 0.95 0.90, 1.01

Weight loss 1.09 1.02, 1.16

Solid tumor without metastasis 1.15 1.09, 1.22

Obesity 0.89 0.83, 0.96

Coagulopathy 1.13 1.02, 1.24

Metastatic cancer 1.42 1.29, 1.55

Lymphoma 1.28 1.15, 1.44

Liver disease 1.23 1.07, 1.43

a Hazard ratios (HR) were adjusted for all other covariates listed in the table. CI�confidence interval,CMG�case-mix group, FIM�Functional Independence Measure.b Variable with significant time interaction; HR reported by time of hospital readmission.

Hospital Readmission and Older Adults With Debility

248 f Physical Therapy Volume 96 Number 2 February 2016 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

with a study of readmission amongpatients with stroke.50 A study ofpatients with hip fracture showed anassociation between higher FIM total rat-ing and lower readmission risk, butmotor and cognition subscales were notdifferentiated.44

The interaction between FIM dischargemotor ratings and time of readmissionwas an interesting finding. Trend in haz-ard ratios for readmission time indicatedthat a higher FIM discharge motor ratingwas more protective for earlier readmis-sions. An increase of 1 point in FIM dis-charge motor rating was associated with2% lower hazard for readmission withinthe first 2 weeks and with a 1% lowerhazard for readmissions at 1 month. Themean FIM discharge motor rating was 3points higher (66.0 versus 63.0) forpatients who did not experience read-mission compared with those who had areadmission within 90 days. A 3-pointhigher FIM discharge motor rating wasassociated with 7% hazard reduction forreadmissions that occurred within 1week following discharge. The FIM dis-charge motor rating was not protective 2months or longer following discharge. Asmore time passed postdischarge, unmea-sured factors such as daily activity andhome or outpatient therapy may haveplayed an important role in a patient’sfunctional status. A report of follow-upinformation collected 80 to 180 daysafter discharge from inpatient rehabilita-tion showed that 61% of patients withdebility received no additional therapy.51

Further research is needed to explorethe effect of follow-up therapy on thetrajectory of functional status after dis-charge from postacute care for patientswith debility.

To our knowledge, this was the firststudy to explore hospital readmission forpatients with debility over a 90-dayperiod. The strengths of this studyincluded use of CMS data, which pro-vided a large national sample andallowed the analysis of numerous factorsobserved in everyday clinical practice.Multiple CMS files provided the opportu-nity to link inpatient rehabilitation vari-ables with acute hospital readmissioninformation. Time-to-event analysisallowed us to observe the distribution of

readmission time and identified variableswith time-dependent hazard.

Our study had several limitations associ-ated with use of administrative dataincluding coding errors and missing data.For variables with known distributions,descriptive statistics were screened toidentify potential data errors. Statisticalscreening can be used to identifyextreme measures, but not categoricalmisclassifications. We were limited tothe variables included in the claims fileswhich frequently lack sensitivity. Thiswas particularly true in areas related tosocial support. For example, marital sta-tus and living status were crude proxiesfor social support. Only comorbidity datarecorded for inpatient admission wasexamined in this study, and outpatientcomorbidity data may have also beenrelevant.52

Our study was limited to Medicare fee-for-service beneficiaries who receivedinpatient rehabilitation for debility andwere discharged to a community setting.Combined criteria for inclusion andexclusion and merging of multiple filesresulted in a sample size that was 35% ofall cases. These cases were generalizableto patients who had Medicare fee-for-

service benefits for old age and com-pleted a typical course of inpatient reha-bilitation for debility following an acutehospitalization. We used only the firsthospital readmission following dis-charge, and patients may have had mul-tiple readmissions during the 90-dayperiod. We examined all-cause readmis-sions and did not differentiate betweenpreventable and unpreventable readmis-sions. The findings were not generaliz-able to the entire Medicare population orto Medicare-managed care. Other post–acute care settings, such as skilled nurs-ing facility, long-term acute care, andhome health, were not examined, andthe influence of selection criteria forpost–acute care setting was unknown.

Despite the above limitations, our find-ings have implications relevant to clinicalpractice and health policy. Patients withdebility who have comorbid conditionsassociated with increased readmissionhazard should be monitored for changesin medical status. For example, vital signsand signs and symptoms of decompensa-tion in patients with heart failure (ie,weight gain, worsening fatigue, dyspnea,or functional decline)53 are important tomonitor and discuss with the interdisci-plinary team. These signs and symptoms

Table 4.Most Common MS-DRG Codes for Hospital Readmission 90 Days After Discharge FromInpatient Rehabilitation for Debilitya

Rank MS-DRG Codes %

1 Heart Failure and Shock (291, 292) 7.6

2 Kidney and Urinary Tract Infections (689, 690) 3.8

3 Nutritional and Miscellaneous Metabolic Disorders (640, 641) 3.6

4 Renal Failure (682, 683) 3.3

5 Simple Pneumonia and Pleurisy (193, 194) 3.2

6 Septicemia Without MV 96� Hours With MCC (871) 3.1

7 Esophagitis, Gastrointestinal, and Miscellaneous Digestive Disorders (391, 392) 3.1

8 Syncope and Collapse (312) 1.7

9 Major Gastrointestinal Disorders and Peritoneal Infections With MCC (371) 1.5

10 Chronic Obstructive Pulmonary Disease With MCC (190) 1.5

11 Gastrointestinal Hemorrhage With MCC (377) 1.3

12 Cardiac Arrhythmia and Conduction Disorders With MCC (308) 1.2

13 Respiratory Infections and Inflammations With MCC (177) 1.1

14 Intracranial Hemorrhage or Cerebral Infarction With MCC (064) 1.1

a MCC�major complication/comorbidity, MS-DRG�Medicare severity diagnosis related groups,MV�mechanical ventilation.

Hospital Readmission and Older Adults With Debility

February 2016 Volume 96 Number 2 Physical Therapy f 249 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

may be recognized days (median�7days) before “overt heart failure decom-pensation.”53 Evaluating discharge motorfunction and comorbidity profile mayhelp the team to identify patients with ahigher probability of hospital readmis-sion. In addition, the timing of readmis-sions indicates that early follow-up maymaximize prevention efforts. These con-siderations are relevant to dischargeplanning and transition of care frominpatient rehabilitation to community.Patients with debility who have lowermotor function and comorbid conditionsassociated with increased readmissionhazard should be monitored for func-tional trajectory and medical stabilityduring inpatient rehabilitation and post-discharge. Optimizing independence lev-els for motor function during inpatientrehabilitation is an important consider-ation for reducing hospital readmissions.Emphasis on higher-intensity resistancetraining and motor task-specific trainingrather than general conditioning (with-out specific exercise parameters) hasbeen recommended for older adults withhospital-associated deconditioning.54

Analysis of individual FIM dischargemotor items in our study indicated thatwalk/wheelchair and stair locomotion,lower body dressing, eating, and boweland bladder control are the most impor-tant functions for readmission hazardamong patients with debility.

These study findings also are applicableto health reform initiatives aimed toreduce hospital readmissions for high-risk Medicare beneficiaries through coor-dination of care across acute and post–acute care settings.55,56 Proposed riskadjustment for readmission rates postdis-charge for inpatient rehabilitationincludes demographics, diagnoses,comorbid conditions, and CMGs, whichincorporate functional motor score.57

Physical therapists contribute functionalstatus information that is relevant toassessment of a patient’s probability forhospital readmission. In collaborationwith an interdisciplinary team, physicaltherapists also provide recommenda-tions for follow-up care after dischargefrom inpatient rehabilitation (ie, homehealth or outpatient services). Patienteducation for monitoring signs andsymptoms associated with change in acu-

ity of comorbid conditions and functionis also a relevant component of dischargeplanning and readiness to transitionfrom inpatient rehabilitation to thecommunity.

In conclusion, Medicare fee-for-servicebeneficiaries who receive inpatient reha-bilitation for debility experienced a highrate of hospital readmission in the sam-ple we studied. Discharge motor func-tional status, several comorbid condi-tions, and marital status were associatedwith readmission. Interactions with timewere found with the following variables:discharge motor functional status, CMGcomorbidity tier, and chronic pulmonarydisease. Future research should buildupon these findings to help developevidence-based guidelines for care tran-sitions in patients with debility.

Dr Galloway, Dr Karmarkar, Dr Graham, DrTan, Dr Raji, and Dr Granger provided con-cept/idea/research design. Dr Galloway, DrTan, Dr Raji, and Dr Ottenbacher providedwriting. Dr Galloway, Dr Karmarkar, and DrTan provided data analysis. Dr Galloway, DrRaji, and Dr Ottenbacher provided projectmanagement. Dr Ottenbacher obtainedfunding support for the project. Dr Raji andDr Granger provided institutional liaisons. Allauthors provided critical review of manu-script before submission.

This study was supported, in part, by grantsfrom the National Institutes of Health (R24HD065702 and R01-HD069443) and theNational Institute on Disability and Rehabil-itation Research (H133G080163).

The FIM instrument is a registered trademarkof the Uniform Data System for MedicalRehabilitation, a division of UB FoundationActivities Inc.

DOI: 10.2522/ptj.20150030

References1 Kortebein P. Rehabilitation for hospital-

associated deconditioning. Am J Phys MedRehabil. 2009;88:66–77.

2 Haley R, Sullivan DH, Granger CV, Korte-bein P. Inpatient rehabilitation outcomesfor older adults with nondebility general-ized weakness. Am J Phys Med Rehabil.2011;90:791–797.

3 Covinsky KE, Palmer RM, Fortinsky RH,et al. Loss of independence in activities ofdaily living in older adults hospitalizedwith medical illnesses: increased vulnera-bility with age. J Am Geriatr Soc. 2003;51:451–458.

4 Sager MA, Franke T, Inouye SK, et al. Func-tional outcomes of acute medical illnessand hospitalization in older persons. ArchIntern Med. 1996;156:645–652.

5 Mahoney JE, Sager MA, Jalaluddin M. Newwalking dependence associated with hos-pitalization for acute medical illness: inci-dence and significance. J Gerontol A BiolSci Med Sci. 1998;53:M307–M312.

6 Centers for Medicare & Medicaid Services.CMS Manual System, Pub 100-02 MedicareBenefit Policy, Transmittal 119. Availableat: https://www.cms.gov/transmittals/downloads/R119BP.pdf. Published Janu-ary 15, 2010. Accessed November 29,2011.

7 UB Foundation Activities Inc. InpatientRehabilitation Facility–Patient AssessmentInstrument (IRF-PAI) Training Manual.Available at: http://www.cms.gov/Medicare/Medicare-Fee-for -Service-P a y m e n t / I n p a t i e n t R e h a b F a c P P S /downloads//irfpaimanual040104.pdf. Pub-lished 2004. Accessed April, 24, 2012.

8 Galloway RV, Granger CV, Karmarkar AM,et al. The Uniform Data System for MedicalRehabilitation report of patients withdebility discharged from inpatient rehabil-itation programs in 2000–2010. Am J PhysMed Rehabil. 2013;92:14–27.

9 Medicare Payment Advisory Commission.Report to the Congress: Medicare PaymentPolicy. Inpatient rehabilitation facility ser-vices. Available at: http://medpac.gov/documents/reports/mar14_entirereport.pdf. Published March 2014. Accessed July6, 2014.

10 Kortebein P, Bopp MM, Granger CV, Sul-livan DH. Outcomes of inpatient rehabili-tation for older adults with debility. Am JPhys Med Rehabil. 2008;87:118–125.

11 Kortebein P, Granger CV, Sullivan DH. Acomparative evaluation of inpatient reha-bilitation for older adults with debility, hipfracture, and myopathy. Arch Phys MedRehabil. 2009;90:934–938.

12 Ottenbacher KJ, Karmarkar AM, GrahamJE, et al. Thirty-day hospital readmissionfollowing discharge from postacute reha-bilitation in fee-for-service Medicarepatients. JAMA. 2014;311:604–614.

13 Centers for Medicare & Medicaid Services,Department of Health and Human Servic-es. Medicare Program; inpatient rehabilita-tion facility prospective payment systemfor federal fiscal year 2015. Fed Regist.2014;79:45871–45936.

14 National Quality Forum. NQF #2502. All-cause unplanned readmission measure for30 days post discharge from inpatientrehabilitation facilities (IRFs). MeasureSteward: Centers for Medicare & MedicaidServices. Available at: http://www.qualityforum.org/QPS/QPSTool.aspx. PublishedDecember 23, 2014. Accessed June 17,2015.

15 Centers for Medicare & Medicaid Services.Bundled Payments for Care Improvement(BPCI) Initiative: General Information.Available at: http://innovation.cms.gov/initiatives/bundled-payments/. Published2013. Accessed January 8, 2015.

Hospital Readmission and Older Adults With Debility

250 f Physical Therapy Volume 96 Number 2 February 2016 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

16 Research Data Assistance Center. MasterBeneficiary Summary File. Available at:http://www.resdac.org/cms-data/files/mbsf. Updated 2012. Accessed October 6,2012.

17 Centers for Medicare & Medicaid Services.Medicare Provider Analysis and Review(MEDPAR) File. Available at: http://www.cms.gov/Research-Statistics-Data-and-Systems/Files-for-Order/IdentifiableDataFiles/MedicareProviderAnalysisandReviewFile.html. Updated December 5,2013. Accessed January 13, 2015.

18 Centers for Medicare & Medicaid Services(CMS) H. Acute Care Hospital InpatientProspective Payment System. Available at:http://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNProducts/downloads/AcutePaymtSysfctsht.pdf. Updated 2012. Ac-cessed December 12, 2012.

19 Keith RA, Granger CV, Hamilton BB, Sher-win FS. The Functional IndependenceMeasure: a new tool for rehabilitation. AdvClin Rehabil. 1987;1:6–18.

20 Linacre JM, Heinemann AW, Wright BD,et al. The structure and stability of theFunctional Independence Measure. ArchPhys Med Rehabil. 1994;75:127–132.

21 Stineman MG, Jette A, Fiedler R, GrangerC. Impairment-specific dimensions withinthe Functional Independence Measure.Arch Phys Med Rehabil. 1997;78:636–643.

22 Ottenbacher KJ, Hsu Y, Granger CV,Fiedler RC. The reliability of the Func-tional Independence Measure: a quantita-tive review. Arch Phys Med Rehabil. 1996;77:1226–1232.

23 Stineman MG, Shea JA, Jette AM, et al. TheFunctional Independence Measure: testsof scaling assumptions, structure, and reli-ability across 20 diverse impairment cate-gories. Arch Phys Med Rehabil. 1996;77:1101–1108.

24 Dodds TA, Martin DP, Stolov WC, Deyo RA.A validation of the Functional IndependenceMeasurement and its performance amongrehabilitation inpatients. Arch Phys MedRehabil. 1993;74:531–536.

25 Reistetter TA, Graham JE, Deutsch A, et al.Utility of functional status for classifyingcommunity versus institutional dischargesafter inpatient rehabilitation for stroke.Arch Phys Med Rehabil. 2010;91:345–350.

26 Corrigan JD, Smith-Knapp K, Granger CV.Validity of the Functional IndependenceMeasure for persons with traumatic braininjury. Arch Phys Med Rehabil. 1997;78:828–834.

27 Granger C, Cotter A, Hamilton B, et al.Functional assessment scales: a study ofpersons with multiple sclerosis. Arch PhysMed Rehabil. 1990;71:870–875.

28 Granger CV, Cotter A, Hamilton D, FiedlerR. Functional assessment scales: a study ofpersons after stroke. Arch Phys Med Reha-bil. 1993;74:133–138.

29 Disler PB, Roy CW, Smith BP. Predictinghours of care needed. Arch Phys MedRehabil. 1993;74:139–143.

30 Carter GM, Buchanan JL, Buntin MB, et al.Executive Summary of Analyses for theInitial Implementation of the InpatientRehabilitation Facility-Prospective Pay-ment System. Santa Monica, CA: RandCorp; 2002.

31 Centers for Medicare & Medicaid Services,Department of Health and Human Servic-es. Medicare Program; changes to theinpatient rehabilitation facility prospec-tive payment system and fiscal year 2004rates. Final rule. Fed Regist. 2003;68:45674–45728.

32 Centers for Medicare & Medicaid Services.Medicare Inpatient Rehabilitation FacilityPPS Data Files. Available at: http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS/Data-Files.html. Updated May 4, 2015.Accessed June 16, 2015.

33 Quan H, Sundararajan V, Halfon P, et al.Coding algorithms for defining comorbidi-ties in ICD-9-CM and ICD-10 administrativedata. Med Care. 2005;43:1130–1139.

34 Elixhauser A, Steiner C, Harris DR, CoffeyRM. Comorbidity measures for use withadministrative data. Med Care. 1998;36:8–27.

35 Shoukri MM, Chaudhary MA. Analysis ofCorrelated Data With SAS and R. 3rd ed.Boca Raton, FL: Chapman & Hall/CRC;2007.

36 Allison PD. Survival Analysis Using SAS:A Practical Guide. 2nd ed. Cary, NC: SASInstitute Inc; 2010.

37 Portney LG, Watkins MP. Correlation. In:Foundations of Clinical Research: Appli-cations to Practice. 3rd ed. Philadelphia,PA: FA Davis Co; 2015:523–538.

38 SAS Institute. SAS/STAT 9.2 User’s Guide:Introduction to Survival Analysis Proce-dures. Available at: http://support.sas.com/documentation/cdl/en/statugsurvival/61761/PDF/default/statugsurvival.pdf. Up-dated 2008. Accessed August 21, 2011.

39 SAS Institute. SAS/STAT 9.2 User’s Guide:The PHREG Procedure. Available at:http://support.sas.com/documentation/cdl/en/statugphreg/61816/PDF/default/statugphreg.pdf. Updated 2008. AccessedAugust 21, 2011.

40 Portney LG, Watkins MP. Multivariateanalysis. In: Foundations of ClinicalResearch: Applications to Practice. 3rd ed.Philadelphia, PA: FA Davis Co; 2015:685–726.

41 Jencks SF, Williams MV, Coleman EA.Rehospitalizations among patients in theMedicare Fee-for-Service Program. N EnglJ Med. 2009;360:1418–1428.

42 Mor V, Intrator O, Feng Z, Grabowski DC.The revolving door of rehospitalizationfrom skilled nursing facilities. Health Aff(Millwood). 2010;29:57–64.

43 Neuman MD, Wirtalla C, Werner RM. Skillednursing facility quality and hospital readmis-sions. JAMA. 2014;312:1542–1551.

44 Ottenbacher KJ, Smith PM, Illilg SB, et al.Hospital readmission of persons with hipfracture following medical rehabilitation.Arch Gerontol Geriatr. 2003;36:15–22.

45 Murphy T, Han L, Allore HG, et al. Treat-ment of death in the analysis of longitudinalstudies of gerontological outcomes. J Geron-tol A Biol Sci Med Sci. 2011;66:109–114.

46 Centers for Medicare & Medicaid Services.Readmissions Reduction Program. Availableat: http://www.cms.gov/Medicare/Medicare -Fee - for -Service -Payment/Acute Inpa t i en tPPS/Readmis s ions -Reduction-Program.html. Published August1, 2012. Accessed October 6, 2012.

47 Fried LP, Tangen CM, Walston J, et al; Car-diovascular Health Study CollaborativeResearch Group. Frailty in older adults:evidence for a phenotype. J Gerontol ABiol Sci Med Sci. 2001;56:M146–M156.

48 Banach M, Bhatia V, Feller MA, et al. Rela-tion of baseline systolic blood pressureand long-term outcomes in ambulatorypatients with chronic mild to moderateheart failure. Am J Cardiol. 2011;107:1208–1214.

49 Gupta PP, Fonarow GC, Horwich TB. Obe-sity and the obesity paradox in heart fail-ure. Can J Cardiol. 2015;31:195–202.

50 Ottenbacher KJ, Graham JE, OttenbacherAJ, et al. Hospital readmission in personswith stroke following postacute inpatientrehabilitation. J Gerontol A Biol Sci MedSci. 2012;67:875–881.

51 Graham JE, Granger CV, Karmarkar AM,et al. The Uniform Data System for MedicalRehabilitation: report of follow-up informa-tion on patients discharged from inpatientrehabilitation programs in 2002–2010. Am J Phys Med Rehabil. 2014;93:231–244.

52 Klabunde CN, Warren JL, Legler JM.Assessing comorbidity using claims data:an overview. Med Care. 2002;40:26–35.

53 Jurgens CY, Goodlin S, Dolansky M, et al.Heart failure management in skilled nurs-ing facilities: a scientific statement fromthe American Heart Association and theHeart Failure Society of America. CircHeart Fail. 2015;8:655–687.

54 Falvey JR, Mangione KK, Stevens-LapsleyJE. Rethinking hospital-associated decon-ditioning: proposed paradigm shift. PhysTher. 2015;95:1307–1315.

55 111th Congress. HR 3590 Patient Protectionand Affordable Care Act. Available at: http://www.hhs.gov/healthcare/rights/law/patient-protection.pdf. Published December 24,2009. Accessed January 1, 2015.

56 Centers for Medicare & Medicaid Services.Community-Based Care Transitions Pro-gram. Available at: http://innovation.cms.g o v / i n i t i a t i v e s / C C T P / ? i t e m I D �CMS1239313. Updated 2015. AccessedJanuary 11, 2015.

57 Centers for Medicare & Medicaid Services.Specifications for the All-Cause UnplannedReadmission Measure for 30 Days Post Dis-charge from Inpatient Rehabilitation Facil-ities. Available at: http://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/IRF-Quality-Reporting/Downloads/DRAFT-Specifications-for-the-Proposed-All-Cause-Unplanned-30-day-Post-IRF-Discharge-Readmission-Measure.pdf. Updated May 2, 2013. AccessedDecember 31, 2014.

Hospital Readmission and Older Adults With Debility

February 2016 Volume 96 Number 2 Physical Therapy f 251 by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from

doi: 10.2522/ptj.20150030Originally published online December 4, 2015

2016; 96:241-251.PHYS THER. Kenneth J. OttenbacherGraham, Alai Tan, Mukaila Raji, Carl V. Granger and Rebecca V. Galloway, Amol M. Karmarkar, James E.

DebilityInpatient Rehabilitation for Older Adults With Hospital Readmission Following Discharge From

References

http://ptjournal.apta.org/content/96/2/241#BIBLfor free at: This article cites 37 articles, 5 of which you can access

Information Subscription http://ptjournal.apta.org/subscriptions/

Permissions and Reprints http://ptjournal.apta.org/site/misc/terms.xhtml

Information for Authors http://ptjournal.apta.org/site/misc/ifora.xhtml

by guest on March 9, 2016http://ptjournal.apta.org/Downloaded from